It won’t be long before the drone technology would be used for delivery of products ordered online by people. Scientists have been working on robotic flyers which have been programed to determine health of the drone while it is inflight so to ensure a safe passage and landing of the packages at their destination.
These drones are miracle workers when it comes to speed! Once the customer has placed the order online, the package would be delivered after just half an hour. Amazon, the retail tycoon, claimed last year in December to start delivery services through the use of drones. However potential threats to drones like shortage of fuel, fast moving winds and erroneous sensors may raise concerns to the safe delivery of the ordered products but to minimize these issues small bots have been designed to assist the drone in its course.
Work has been in progress at Cambridge, MIT (Massachusetts Institute of Technology). An aeronautical engineer there, Ali-akbar Agha-mohammadi provided a statement saying, “”With something like package delivery, which needs to be done [consistently] over hours, you need to take into account the health of the system”. Along with his colleagues, Agha mohammadi has been able to generate an algorithm that makes the drones self-sufficient enough to manage, monitor and protect its own sensors and machinery. The drone is designed to be an intelligent machine as it can change course to find the closest refueling/charging station in case it encounters fuel shortage. This is just one example of how a drone can tackle problems that come its way.
The drone also makes calculative decisions about which route or course to take in order to evade collisions before taking off for its destined area as mentioned by the team at MIT. In order to put the bots to test, researchers made a comparison of them. One set of bots had been installed with the health monitoring system where as the other set was not. Multiple deliveries were to be made by the bots which is why the drones were exposed to an environment that was closest to reality. The scientists reported lesser breakdowns and glitches from the bots which were installed with the health monitoring system however both types were able to reach their destinations with the given packages.
The Markov decision process, normally used by automatic machinery in taking decisions like during navigation, provides an array of possibilities where each probability forks into further applicable options. The task of this vigorous process is to find the path or option that poses least amount of risk and can be executed. However the researchers commented that sadly this model is not so applicable in reality because a slight change in weather can change the entire plan, messing up the measurements. So they approached the problem in a somewhat different way. This approach includes the older version of making decisions but the only difference is that each possibility would branch out into a certain executable action. This large amount of workable data needs to be computed and processed which can be achieved by two ways.
Researchers found these ways to make the data much easier to manage so firstly vehicle planning helps in tracking the location of the drone whereas mission-level planning would monitor the health and condition of the drone equipment and parts. Vehicle planning is more specific and possibilities are narrowed down early on by the drone so that more energy and processing power can be spent while the drone is inflight to manage mission level planning.
Scientists wish to send real drones for a test flight, carrying delivery packages, to check the performance of health-monitoring system installed in the drone. The team at MIT is hard at work so they can present their observations and findings in Chicago, this September at the IEEE/RSJ International Conference on Intelligent Robots and Systems.